PLI error for pdb to pqr conversion
Opened this issue · 1 comments
zruan commented
Hello developers,
I'm trying to use the ESP_DNN to calculate the ESP of a ligand molecule. The input pdb file looks like this:
HETATM 1 N UNK C 601 122.444 122.637 75.742 0.00 0.00 N
HETATM 2 N UNK C 601 125.882 124.910 87.137 0.00 0.00 N
HETATM 3 C UNK C 601 123.711 123.036 79.714 0.00 0.00 C
HETATM 4 C UNK C 601 124.672 124.497 83.166 0.00 0.00 C
HETATM 5 C UNK C 601 124.651 122.923 78.689 0.00 0.00 C
HETATM 6 C UNK C 601 123.715 124.539 84.168 0.00 0.00 C
HETATM 7 C UNK C 601 124.107 123.177 81.109 0.00 0.00 C
HETATM 8 C UNK C 601 124.290 124.354 81.750 0.00 0.00 C
HETATM 9 C UNK C 601 122.349 123.013 79.415 0.00 0.00 C
HETATM 10 C UNK C 601 126.024 124.589 83.460 0.00 0.00 C
HETATM 11 C UNK C 601 124.229 122.790 77.367 0.00 0.00 C
HETATM 12 C UNK C 601 124.119 124.677 85.496 0.00 0.00 C
HETATM 13 C UNK C 601 122.867 122.769 77.069 0.00 0.00 C
HETATM 14 C UNK C 601 125.476 124.770 85.806 0.00 0.00 C
HETATM 15 C UNK C 601 121.926 122.880 78.093 0.00 0.00 C
HETATM 16 C UNK C 601 126.428 124.727 84.788 0.00 0.00 C
HETATM 17 C UNK C 601 122.310 123.672 75.190 0.00 0.00 C
HETATM 18 C UNK C 601 126.070 123.878 87.678 0.00 0.00 C
HETATM 19 O UNK C 601 126.700 124.480 79.466 0.00 0.00 O
HETATM 20 O UNK C 601 121.762 122.882 83.355 0.00 0.00 O
HETATM 21 O UNK C 601 126.620 122.096 80.166 0.00 0.00 O
HETATM 22 O UNK C 601 127.039 122.694 77.744 0.00 0.00 O
HETATM 23 O UNK C 601 121.289 124.599 85.089 0.00 0.00 O
HETATM 24 O UNK C 601 121.737 125.298 82.700 0.00 0.00 O
HETATM 25 S UNK C 601 126.374 122.945 79.011 0.00 0.00 S
HETATM 26 S UNK C 601 121.999 124.426 83.833 0.00 0.00 S
HETATM 27 S UNK C 601 122.122 125.054 74.420 0.00 0.00 S
HETATM 28 S UNK C 601 126.332 122.498 88.432 0.00 0.00 S
HETATM 29 H UNK C 601 124.233 122.243 81.651 0.00 0.00 H
HETATM 30 H UNK C 601 124.163 125.291 81.212 0.00 0.00 H
HETATM 31 H UNK C 601 121.601 123.098 80.201 0.00 0.00 H
HETATM 32 H UNK C 601 126.780 124.556 82.680 0.00 0.00 H
HETATM 33 H UNK C 601 124.941 122.702 76.550 0.00 0.00 H
HETATM 34 H UNK C 601 123.396 124.714 86.307 0.00 0.00 H
HETATM 35 H UNK C 601 120.863 122.862 77.872 0.00 0.00 H
HETATM 36 H UNK C 601 127.489 124.800 85.017 0.00 0.00 H
HETATM 37 H UNK C 601 126.522 124.698 80.408 0.00 0.00 H
HETATM 38 H UNK C 601 121.993 122.193 84.021 0.00 0.00 H
CONECT 1 13 17
CONECT 2 14 18
CONECT 3 5 7 9
CONECT 4 6 8 10
CONECT 5 3 11 25
CONECT 6 4 12 26
CONECT 7 3 8 29
CONECT 8 4 7 30
CONECT 9 3 15 31
CONECT 10 4 16 32
CONECT 11 5 13 33
CONECT 12 6 14 34
CONECT 13 1 11 15
CONECT 14 2 12 16
CONECT 15 9 13 35
CONECT 16 10 14 36
CONECT 17 1 27
CONECT 18 2 28
CONECT 19 37 25
CONECT 20 38 26
CONECT 21 25
CONECT 22 25
CONECT 23 26
CONECT 24 26
CONECT 25 5 19 21 22
CONECT 26 6 20 23 24
CONECT 27 17
CONECT 28 18
CONECT 29 7
CONECT 30 8
CONECT 31 9
CONECT 32 10
CONECT 33 11
CONECT 34 12
CONECT 35 15
CONECT 36 16
CONECT 37 19
CONECT 38 20
END
After running python -m esp_dnn.predict -m ligand -i examples/px1_drugs/, I got the following error:
[predict:INFO] Ruuning in ligand mode
[predict:INFO] Input dir is /legacy/storage/projects/lu/Zheng/software/ESP_DNN/examples/px1_drugs
[predict:INFO] Output dir is /legacy/storage/projects/lu/Zheng/software/ESP_DNN/examples/px1_drugs
[predict:INFO] Loading the trained model from /legacy/storage/projects/lu/Zheng/software/ESP_DNN/esp_dnn/model directory
Using TensorFlow backend.
2023-04-09 17:47:31.377224: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA
[predict:INFO] /legacy/storage/projects/lu/Zheng/software/ESP_DNN/examples/px1_drugs/dids.pdb -> /legacy/storage/projects/lu/Zheng/software/ESP_DNN/examples/px1_drugs/dids.pdb.pqr
[predict:ERROR] Error converting /legacy/storage/projects/lu/Zheng/software/ESP_DNN/examples/px1_drugs/dids.pdb file
[predict:ERROR] PLI failed to create pqr file
PLI Error:
Fatal error:
set_fgroup_type: matched fgroup type does not match hydrogens for atom ligand:7
The error message is not very informative to me. Do you know what might be wrong in this case?
Thank you so much!
fredludlow commented
I suspect it's the odd geometry around that central double bond? (atoms 7 and 8). It might help to have an input file with double bonds specified using the (unofficial) multiple CONECT record format, e.g. there's a double bond between 7 and 8 so it would look like:
CONECT 7 3 8 8 29
CONECT 8 4 7 7 30
